10 version = "0-unstable-2024-07-29";
11 nativeBuildInputs = [ unzip ];
14 description = "NLTK Data";
15 homepage = "https://github.com/nltk/nltk_data";
16 license = licenses.asl20;
17 platforms = platforms.all;
18 maintainers = with maintainers; [ happysalada ];
28 src = fetchFromGitHub {
31 rev = "cfe82914f3c2d24363687f1db3b05e8b9f687e2b";
33 sparseCheckout = [ "packages/${location}/${pname}.zip" ];
36 stdenvNoCC.mkDerivation (
40 inherit (base) version;
45 unzip ${src}/packages/${location}/${pname}.zip
46 mkdir -p $out/${location}
47 cp -R ${pname}/ $out/${location}
54 lib.makeScope newScope (self: {
55 punkt = makeNltkDataPackage {
57 location = "tokenizers";
58 hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
60 punkt_tab = makeNltkDataPackage {
62 location = "tokenizers";
63 hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
65 averaged_perceptron_tagger = makeNltkDataPackage {
66 pname = "averaged_perceptron_tagger";
68 hash = "sha256-tl3Cn2okhBkUtTXvAmFRx72Brez6iTGRdmFTwFmpk3M=";
70 snowball_data = makeNltkDataPackage {
71 pname = "snowball_data";
72 location = "stemmers";
73 hash = "sha256-mNefwOPVJGz9kXV3LV4DuV7FJpNir/Nwg4ujd0CogEk=";
75 stopwords = makeNltkDataPackage {
78 hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
80 wordnet = makeNltkDataPackage {
83 hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";